Unmasking biases and navigating pitfalls in the ophthalmic artificial intelligence lifecycle: A narrative review

Over the past 2 decades, exponential growth in data availability, computational power, and newly available modeling techniques has led to an expansion in interest, investment, and research in Artificial Intelligence (AI) applications. Ophthalmology is one of many fields that seek to benefit from AI given the advent of telemedicine screening programs and the use of ancillary imaging. However, before AI can be widely deployed, further work must be done to avoid the pitfalls within the AI lifecycle. This review article breaks down the AI lifecycle into seven steps—data collection; defining the model task; data preprocessing and labeling; model development; model evaluation and validation; deployment; and finally, post-deployment evaluation, monitoring, and system recalibration—and delves into the risks for harm at each step and strategies for mitigating them.Author summary: In recent years, the surge in data availability, computational power, and AI techniques has sparked interest in using AI in fields like ophthalmology. However, before widespread AI deployment can happen, we must carefully navigate its lifecycle, comprising 7 key steps: data collection, task definition, data preparation, model development, evaluation, deployment, and post-deployment monitoring. This review article stands out by identifying potential pitfalls at each stage and offering actionable strategies to address them. Our article serves as a guide for harnessing AI effectively and safely in ophthalmology and related fields.